Canadian news publishers ask Meta to scrap proposed AI-training terms, citing worries about data use

Canadian news publishers are pushing back on proposed changes to Meta’s terms of service that they say could broaden how their content is used to train artificial intelligence models. The concern is shared by publishers beyond Canada, who view the change as part of an increasingly tense debate over content licensing and AI data rights.

In Canada, the industry association New Media Canada has publicly called for Meta to drop or revise the new terms, according to reporting carried by Yahoo Finance. The publishers’ core argument is that the language would allow Meta to use content more broadly for training and related purposes, raising the prospect that outlets could be effectively opted into AI training without explicit permission.

The request is arriving at a moment when governments, regulators, and media companies are working through questions that did not exist in earlier AI rollouts. Publishers want clearer boundaries on whether user and editorial content posted online can be reused for training, and under what consent or compensation framework.

Meta, for its part, has been actively expanding its AI capabilities across products, from ranking and recommendations to generative AI features. However, the reported Canadian objection is not focused on Meta’s general AI strategy, but specifically on whether its terms of service changes shift control over third-party content.

Public information about exactly what the revised terms would permit is limited in the reporting described here. The article notes the publishers’ fear that the proposed language could allow Meta to use content to train AI systems, but it does not provide a full, clause-by-clause breakdown of the changes in the information available for this draft.

In the absence of more detailed disclosure in the cited reporting, it remains unclear how Meta’s proposal would apply in practice. For example, it is not established in the material reviewed here whether publishers could opt out, whether the proposal would cover all content or only certain categories, or whether any compensation or licensing mechanisms would be offered.

Still, the pressure campaign indicates that publishers are increasingly treating AI training as a business model question, not only a technical one. If major platforms move forward with terms that publishers view as expansive, outlets may respond with new policies, contractual demands, or enforcement strategies aimed at restricting AI use.

What to watch next is whether Meta clarifies the scope of the proposed terms and whether it engages directly with Canadian publishers or industry groups. Equally important will be whether regulators take up the underlying question of consent and content reuse for AI training, especially as media organizations seek enforceable guardrails.